Reply to Topical Editor Decision: Reconsider after major revisions Dear members of the Editorial Board, First of all, we would like to thank you for reconsidering our manuscript for publication. Please find enclosed our detailed reply to the Reviewer #2’s comments on the revised manuscript, the changes in the revised manuscript and the last version of the manuscript. In particular, we addressed the main issue of the reviewer and now explicitly indicate in the abstract and the conclusion that our solution is not unique. We hope that all issues have been convincingly addressed and that the revised manuscript can be accepted for publication. Yours sincerely On behalf of all co-authors, Ershad Gholamrezaie Reply to second review (Reviewer#2) of the revised manuscript General remark: The authors have acknowledgedly included the available shipboard gravity data in their study. They arrive at similar conclusions as in their original version, e. g. that density contrasts are required below the Ganos and the Tuzla bends to fit the gravity field. This conclusion makes sense, as it is consistent with wide-angle reflection-refraction (WARR) seismic data suggesting abrupt uplift of crust and Moho reflectors below both bends. WARR, however, mainly provide constraints along East-West, 2D-profiles (see Figure 3 in Bécel et al, 2009, Tectonophysics 467 , 1–21 ), but not along North-South Profiles. As a result, the 3D geometry of the deep crustal structure is very difficult to assess, based on WARR. In the Eastern Sea of Marmara, the inferred highdensity intrusion is consistent with the presence of a highly magnetized body more or less below the Tuzla bend. In contrast, in the Western Sea of Marmara, there is no evidence for any 3D, high-density crustal body below the Ganos bend, neither from multi-channel reflection seismics, neither from receiver functions along a 650 km transect crossing Western Anatolia at 28◦ E longitude (Karabulut et al, Geophys. J. Int, 194, 450–464). This raises questions on the reality of the inferred High sensity body below the Ganos bend. The workflow adopted ∼by the authors is a classical one, carefully described in the paper. It consists in (1) setting up an,initial density model based on Hergert’s et al previous studies ; (2) calculating the gravity response between the modelled and the observed gravity; (3) modifying the initial model by introducing additional density variations to obtain the best fitting model. The authors end up with ONE solution, which is substantially different from the picture proposed by Kende’s et al (2017). The limitations of the available seismic surveys in the Sea of Marmara is one of our motivations for the 3D gravity modelling as provides another constraint on the density distribution. The bathymetry of the Sea of Marmara especially within the North Marmara Trough (NMT) raises difficulties for seismic data processing and interpreting. In particular, the sharp scarps of the NMT cause strong diffraction that may affect the first Fresnel zone (≈ lateral resolution) during the seismic processing steps and increase the chance of data loss due to the required noise reduction. If the high-density bodies exist as dome-shape intrusive mafic bodies, as suggested by our gravity modelling, they also can diffract the seismic waves. This situation makes the high-density bodies more difficult to detect by seismic surveys. It becomes even more difficult in greater depth. Across the Sea of Marmara, most of the multi-channel reflection seismic surveys aimed to image the sediment layers and faults within the sedimentary basins. In summary, high-resolution 3D seismic survey might be an effective method to clearly image the crustal structures beneath the Princes’ Islands segment and the Ganos Bend, but at the greater depth they have limits and gravity models are a complementary approach to resolve the depth distribution in particular in greater depth. In addition, the study of Karabulut et al. (2013) aimed to image the Moho depth by the computation of receiver functions and performing common conversion point (CCP) depth migration of the P to S converted phase along the ~650 km long NS profile in the western Anatolia. The locations of the stations in the area of the Marmara Sea do not correlate spatially with the location of the modelled thick high-density body below the Ganos Bend. Therefore, the P-wave velocity and the Vp/Vs ratio that are calculated beneath each station cannot detect the high-density body. A tomography study of Yamamoto et al. (2017), however, indicates a zone of higher S-wave velocity and slightly higher P-wave velocity at about 20 km depth b.s.l. beneath the Ganos Bend in the area where the western high- density body cuts the boundary between the upper and the lower crystalline crust. As our work does not contribute to this topic in terms of providing new evidence based on seismic methods we leave these considerations as part of the open review-discussion and did not include them in the manuscript. Finally, we do not agree with the reviewer that we end up with one solution. Actually, we end up with one concept that indicates lateral density heterogeneities beneath the bent segments of the Main Marmara Fault. Considering the same dataset as used in Kende et al., 2017 (Improved–TOPEX), we present two endmember solutions for the configuration of high-density bodies with different density-geometry configurations (and add more details in the supplementary information) that illustrate the spectrum of possible solutions. However, we do agree with the reviewer that except for the tomography results of Yamamoto et al. (2017) there is no evidence for the crustal high-density body below the Ganos bend. This matter and the non-uniqueness of gravity model solutions, in general, leaves us with a question mark about the existence of the western high-density body. We discuss this in section “5.4. Model limitations” and in addition, we added the following paragraph to this section: Page 17, Line 10: “While the aeromagnetic maps (Ates et al., 2003; 2008) indicate a clear positive anomaly (indicative for a mafic body at depth) beneath the Çınarcık Basin that spatially correlates with the eastern high- density bodies, there are no such indications for the western high-density body beneath the Ganos Bend. Considering the non-uniqueness of solutions in potential field modelling, other possible solutions based on different initial models should also be contemplated beneath the Ganos Bend (e.g. Kende et al., 2017; see Fig. S4 and S5 in the Supplement).” Recommendation: My recommendation is « major revisions ». The authors should insist that the solution they propose is not unique. They should discuss the possibility of fitting other solutions below the Ganos bend (e.g. Kende et al, 2017), when other initial models are choosen. The non-uniqueness issue should be explicitly mentionned in the abstract and in the conclusion. We would like to thank the reviewer for revising his/her recommendation towards “major revisions” instead of “reject”. Regarding the discussion of “possibility of fitting other solutions below the Ganos bend”, please see our answer above to the general remark. As recommended by the reviewer, the non-uniqueness issue has been added as follows in the abstract and conclusion sections of the manuscript: Abstract. Page 1, Line 15: “Considering the two different datasets and the general non-uniqueness in potential field modelling, we suggest three possible “endmember” solutions that are all consistent with the observed gravity field and illustrate the spectrum of possible solutions.” Conclusion. Page 19, Line 18: “(5) The configurations of the high-density bodies are exclusively based on 3D forward gravity modelling, a method characterized by an inherent non-uniqueness of the solutions. Only for the eastern bend, seismic and magnetic data support the presence of a deep high-density body, whereas for the western bend such indications are missing. Therefore, further geophysical observations are required to further constrain the detailed density-geometry configuration of these bodies.” Other remark: Please note that in my previous review, I never claimed that the authors have used the « wrong » dataset. My concern was that the authors may have used a dataset that could not be « adapted » to the practical purpose of their study. There is a nuance between « wrong » and « not adapted ». Second, the question is not to check the quality of the « published and downloadable gravity data », but to check that the dataset they use is suitable for the objectives of their study. We acknowledge the reviewer’s concerns regarding the datasets that we had used in our first submitted manuscript and apologize for the misunderstanding related to “wrong” and “not adapted”. It was a good experience to see how an open discussion could be helpful and constructive. 3D Crustal Density Model of the Marmara Sea Ershad Gholamrezaie1,2, Magdalena Scheck-Wenderoth1,3, Judith Sippel1, Oliver Heidbach1, and Manfred R. Strecker2 1Helmholtz Centre Potsdam–GFZ German Research Centre for Geosciences, Potsdam, Germany 5 2Institute of Earth and Environmental Science, University of Potsdam, Germany 3Faculty of Georesources and Material Engineering, RWTH Aachen, Aachen, Germany Correspondence to: Ershad Gholamrezaie ([email protected]) Abstract. The Sea of Marmara, in Northwest Turkey, is a transition zone where the dextral North Anatolian Fault Zone (NAFZ) propagates westward from the Anatolian plate to the Aegean plate. The area is of interest in the context of seismic 10 hazard in the vicinity of Istanbul, a metropolitan area with about 15 million inhabitants. Geophysical observations indicate that the crust is heterogeneous beneath the Marmara Basin, but a detailed characterization of the crustal heterogeneities is still missing. To assess if and how crustal heterogeneities are related to the NAFZ segmentation below the Marmara Sea, we develop a new crustal-scale 3D density modelmodels which integratesintegrate geological and seismological data and isthat are additionally constrained by 3D gravity modelling considering.